@@ -91,8 +91,9 @@ def __init__(self):
9191 ["(?<=[^a-zA-Z])at([^a-zA-Z])" , " @ " ], # replace at with @
9292 ["−" , "-" ], # unify dash (long, short) symbols
9393 ["·" , "*" ], # unify multiplication symbols
94- ["° F" , " °F" ], # pint gets confused (degree farad)
95- ["° C" , " °C" ], # pint gets confused
94+ ['ºC' , 'degC' ], # pint gets confused
95+ ["° F" , "degF" ], # pint gets confused (degree farad)
96+ ["° C" , "degC" ], # pint gets confused
9697 ["°F" , "degF" ], # eliminates issue with capitalization step
9798 ["°C" , "degC" ], # eliminates issue with capitalization step
9899 ["(?<=[0-9]{1})[ ]{0,1}X[ ]{0,1}(?=[0-9]{1})" , "*" ], # unify multiplication symbols
@@ -101,6 +102,7 @@ def __init__(self):
101102 [r"\]" , ")" ], # make all brackets parenthesis
102103 ["^.*={1}" , "" ], # delete everything in front of equal
103104 ["^.*:{1}" , "" ], # delete everything in front of collen
105+ ["±[ ]*[1-9.]+[ ]*" , "" ], # delete the +- range
104106 ["( to )" , "-" ], # unify how range are represented
105107 ["(?<=[a-zA-Z])-(?=[a-zA-Z])" , " " ], # turn dashes between text into spaces so dictionary can remove
106108 ["mm Hg" , "mmHg" ], # pint gets confused
0 commit comments